We proposed research to devise and improve pharmacokinetic (PK) and pharmacodynamic (PD) data-analyiss methods, with special emphasis on the latter. These mehtods provide parameter estimates, confidence regions for parameters, and tests of hypotheses about parameter or the goodness of fit of alternative models. Given the realities of clinicla pharmacologicla data concerning drug concentrations and drug actions, the data analysis methods we devise and test must be able to analyze data with varying characteristics: (1) Experimental-type data (many observations per individual; few individuals) or observational-type data arising from patient care or as add-on data to ongoing clinical trails (few observations per individual; many individuals). (2) Data viewed as arising from an individual or as arising from a population. The distinction is that from the former viewpoint, the focus is the PK/PD behavior of the paticular individual, while from the latter viewpoint, the focus is population mean PK/PD behavior, inter-individual variability and residual intra-individual an measurement error variability. (3) Continuous-valued (cardinal) response, such as drug concentrations, or blood pressure or categories (often ordinal) responses, such as "response" vs "no response". (4) Univariate responses, such as plasma concentrations, or multivariate responses, such ase concentrations in both plasma and urine, plus pharmocologic effects. This category also logically includes mulitivariate responses, some elements of which ar continuous-valued and some categorical. For our research to have meaning, it must be available for use by others. We have implemented our methods as exportable computer programs, and have distributed them to users around the world. Furter such implementation is planned in this renewal. Although we do not request direct grant support to do so, we plan also to apply our methods to real-world data. This supplies needed insight and impetus for our research.